Why are some technologies adopted quickly, while others are not? Why are some surgical procedures adopted wholeheartedly and others simply dismissed? Why does this vary from place to place? In this month’s Editor’s Spotlight/Take 5, Daniel C. Austin MD, MS and his team from Dartmouth-Hitchcock Medical Center, already known for their work on surgical variation , show profound differences in adoption of arthroscopic repair of rotator cuff tears between hospital referral regions across the United States . Why might this be so?
Providers may surmise that a patient in Paris, TX, USA and one in Rome, GA, USA (to borrow half a line from a country song) both will receive the same surgical recommendation if they present with the same clinical problem. But while variation is typically quite low for straightforward problems—surgery is appropriate for almost all patients with hip fractures—the degree of variation for other diagnoses is surprising. For example, coils are used to treat cerebral aneurysms in 99% of patients covered by Medicare in Tacoma, WA, USA, while less than 40% of patients treated for unruptured aneurysms received coiling in Modesto, CA, USA, Madison, WI, USA and Manchester, NH, USA .
What about when new procedures or approaches are introduced? Presumably, we’re all reading the same studies and getting the same information on the risks and benefits of a new technique. It’s reasonable to think that adoption would occur with similar patterns, and over the same amount of time.
But this is almost never the case.
There are notable differences in how individuals approach technological novelty. The first systematic assessment on how to incorporate new practice into daily life dates back to the early 20th century. Ryan and Gross  looked at how Iowa farmers used hybrid seed corn and found features that resonate even today: There was a time lag of about 5 years between first knowledge and first adoption; those who tried the approach earliest were somewhat tentative in incorporating new seed; late adopters were more likely to carry out a full conversion, and “almost all had heard of the new trait before more than a handful were planting it” . Farmers fell into several categories: Innovators, early adopters, early majority, late majority, and laggards—familiar territory for surgeons.
In the early 1960s, Everett Rogers developed the theory of diffusion of innovations, and proposed four elements that influence the spread of a new idea: (1) The innovation itself, (2) communication channels, (3) time, and (4) a social system . Although Rogers did not write specifically about medical or surgical techniques, his model resonates with providers trying out new forms of therapy.
Instead of determining whether arthroscopic rotator cuff repair is better, worse, or no different from traditional open techniques, the authors of this month’s Spotlight article examined the variables affecting how quickly the procedure was adopted in specific regions . Payers and policymakers have much of the same access to published studies as surgeons, so ascribing differences in adoption by region to insurance coverage discrepancies does not adequately explain the finding. Revisiting the theory of diffusion, we see no differences in how arthroscopic cuff repair is done, how it’s communicated, or the amount of time since its introduction. Rather, the differences appear to stem from social characteristics both of surgeons and patients.
Why is this important for all orthopaedic surgeons? Surgeons and their patients want to believe—and patients may reasonably expect—that the reasons a surgeon will recommend a treatment are rooted in the evidence, such that those recommendations won’t vary much from place to place. But surgeons’ recommendations vary. As with the Iowa farmers adopting new practices, surgeons are influenced by what those nearby are doing, by the perceived level of resulting success or failure, and by the reputation of those early adopters. And, importantly, the study in this month’s CORR®  suggests that patients may heavily influence this process. While in principle, this is not a bad thing—surgeons should always consider patients’ preferences—the reasons may have more to do with systematic differences. Austin and colleagues  found that regions of lower levels of education and healthcare spending generally pursued adoption of new technology more slowly. Perhaps surprisingly, though, regions with academic centers were also slower to adopt arthroscopic repair—so there’s more here than demographics!
In short, the message often varies from place to place.
With that, let’s talk with Daniel C. Austin MD, MS, lead author of “Mapping the Diffusion of Technology in Orthopaedic Surgery: Understanding the Spread of Arthroscopic Rotator Cuff Repair in the United States”
Take 5 Interview with Daniel C. Austin MD, MS, lead author of “Mapping the Diffusion of Technology in Orthopaedic Surgery: Understanding the Spread of Arthroscopic Rotator Cuff Repair in the United States”
Paul A. Manner MD:Congratulations on publishing this exciting paper. Your group has done a great deal of work on variation in care of surgical conditions. For example, you found little variation in hip fracture treatment, but considerable variation in approaches to treatment of carotid stenosis . Is there a “sweet spot?” How much variation is acceptable?
Daniel C. Austin MD, MS: We have studied this very question and published our thoughts in CORR back in 2009 . And a decade later, variation remains an integral part of medical care and in many contexts can be viewed positively—we want our physicians to carefully consider what is best for our specific situation and not to simply give us the exact same treatment as everybody else. This variation, which appropriately responds to differences in the health needs and preferences of the population, can be considered “warranted variation.” Unwarranted variation refers to the differences in rates of treatments and procedures that cannot be attributed to the underlying needs and preferences of the population. The rates of unwarranted variation tend to be high when there is diagnostic uncertainty about the condition, a lack of consensus regarding the most appropriate treatment, and particularly when there is a large spectrum of disease with a large reservoir of mild or even subclinical disease and a lack of clear criteria for when to most appropriately intervene. While there is no “right rate” of variation (and no reason to believe that the appropriate level of variation is the same across all conditions and interventions), the general approach to finding the most appropriate rate for a given procedure in a given population involves patients receiving unbiased information on the pros and cons of the available treatment options. Patients can then apply their personal preferences through a formalized process of shared-decision making.
In terms of this project, it has been previously suggested that innovation requires variation in practice and may be at odds with the standardization of evidence-based medicine . The high degree of variation in the adoption of arthroscopic rotator cuff repair we observed in 2006 may be a normal and necessary part of diffusion. However, the ongoing variation we observed in the adoption of arthroscopic rotator cuff repair in 2014 was likely due in part to its general equivalence to open repair in terms of long-term outcomes. While it is difficult to quantify an acceptable degree of variation, high variation procedures such as rotator cuff repair, highlight the need to further high-quality research and clinical practice guidelines to create better treatment algorithms.
Dr. Manner:One of the things we know from previous work is that the strongest predictor for spine surgery is the presence of a spine surgeon [3, 8]. It’s possible that a big change in use of a given technique for a particular location is that a new surgeon entered the area, or a previously established one left. Might the changes you see reflect this evolution?
Dr. Austin: In our analysis of hospital referral regions, it is possible that big changes could be due to surgeons entering or leaving an area. This is known as “supply-sensitive care,” and highlights the role that surgeons can have in affecting utilization or adoption of a given procedure . However, several regions such as Provo, UT, USA stayed at the top from 2006 to 2014. Also, many broader trends are present when looking at the nationwide maps, and the mountain west jumps out as an entire region that adopted arthroscopic rotator cuff repair relatively earlier. On a related note, the rate of rotator cuff repair was not related to early adoption, suggesting that areas that did more repairs did not necessarily adopt arthroscopic repairs earlier and that utilization and innovation are not synonymous.
Dr. Manner:It’s fashionable to criticize insurance companies for not paying for new procedures, but it seems like they often have a point. We have the example of knee arthroscopy for arthritis , which was demonstrably ineffective—surgeons didn’t change practice until Medicare started denying reimbursement. What’s the role of payers/insurers here?
Dr. Austin: It seems reasonable that the newest and most experimental procedures would not be covered by insurance companies since they are unproven, and highlights the important role that grant and industry funding may have in nurturing innovation in its infancy. However, payers have not always acted with complete transparency when approving or denying interventions. This highlights the need for impartial (non-biased) reviews of current evidence, which is the gap that evidence based clinical practice guidelines are designed to fill. Once reasonable evidence suggests that a given innovation is equal or better than the status quo, without an excessive increase in cost, it is necessary for insurance companies to reimburse appropriately.
Dr. Manner:Is there a danger of looking too closely? As an example, the Chico, CA, USA metropolitan area has a population of about 200,000, which means that about 300 rotator cuff repairs would be done each year. Similarly, Altoona, PA, USA has population of about 130,000, which might correspond to about 200 rotator cuff repairs. A low (or high) rate might simply reflect the dominance of a single rather busy shoulder surgeon, or a lack of anyone doing these.
Dr. Austin: This is a great point. Relatively small regions can be substantially affected by the change in practice of a single surgeon. Within the top and bottom 20 regions listed in our paper, there are several lower-population areas highlighting this issue. The methodology we used did censor the lowest population areas so that there was a minimum number of procedures (26 per year) performed to lower the risk of this type of problem. It is important to consider broader trends present on the nationwide maps to better understand variation on a slightly larger scale. Importantly, our regression modeling was adjusted for the population of a given region and allowed us to evaluate nationwide trends.
Dr. Manner:Finally (again using Chico and Altoona), there are major differences in demographics. Chico is expanding rapidly (hence, more attractive to a surgeon looking to grow a practice) while Altoona is shrinking (and presumably less attractive to that same surgeon). Is it possible that big differences in small regions may reflect local differences and not anything systemic?
Dr. Austin: It is possible that additional confounders could be underlying the trends that we observed. To expand upon your example, it is possible that more educated areas with better schools would be more attractive to younger, presumably more innovative, surgeons. Nevertheless, we attempted to capture many socioeconomic and healthcare variables in our models to understand what overlying trends were present. Ultimately, the context in which a surgeon practices likely influences their motivation to innovate, and it is plausible that some of the factors identified in our study, including higher-education and increased healthcare spending, could be associated with areas that innovate more.